Convolutional Neural Networks

Project: Write an Algorithm for a Dog Identification App


In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the Jupyter Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.

In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this Jupyter notebook.


Why We're Here

In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).

Sample Dog Output

In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!

The Road Ahead

We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.

  • Step 0: Import Datasets
  • Step 1: Detect Humans
  • Step 2: Detect Dogs
  • Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
  • Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)
  • Step 5: Write your Algorithm
  • Step 6: Test Your Algorithm

Step 0: Import Datasets

Make sure that you've downloaded the required human and dog datasets:

Note: if you are using the Udacity workspace, you DO NOT need to re-download these - they can be found in the /data folder as noted in the cell below.

  • Download the dog dataset. Unzip the folder and place it in this project's home directory, at the location /dog_images.

  • Download the human dataset. Unzip the folder and place it in the home directory, at location /lfw.

Note: If you are using a Windows machine, you are encouraged to use 7zip to extract the folder.

In the code cell below, we save the file paths for both the human (LFW) dataset and dog dataset in the numpy arrays human_files and dog_files.

In [1]:
import numpy as np
from glob import glob

# load filenames for human and dog images
human_files = np.array(glob("/data/lfw/*/*"))
dog_files = np.array(glob("/data/dog_images/*/*/*"))

# print number of images in each dataset
print('There are %d total human images.' % len(human_files))
print('There are %d total dog images.' % len(dog_files))
There are 13233 total human images.
There are 8351 total dog images.

Step 1: Detect Humans

In this section, we use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images.

OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory. In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.

In [2]:
import cv2                
import matplotlib.pyplot as plt                        
%matplotlib inline                               

# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')

# load color (BGR) image
img = cv2.imread(human_files[0])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# find faces in image
faces = face_cascade.detectMultiScale(gray)

# print number of faces detected in the image
print('Number of faces detected:', len(faces))

# get bounding box for each detected face
for (x,y,w,h) in faces:
    # add bounding box to color image
    cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
    
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()
Number of faces detected: 1

Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.

Write a Human Face Detector

We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.

In [3]:
# returns "True" if face is detected in image stored at img_path
def face_detector(img_path):
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray)
    return len(faces) > 0

(IMPLEMENTATION) Assess the Human Face Detector

Question 1: Use the code cell below to test the performance of the face_detector function.

  • What percentage of the first 100 images in human_files have a detected human face?
  • What percentage of the first 100 images in dog_files have a detected human face?

Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.

Answer: (You can print out your results and/or write your percentages in this cell)

In [4]:
from tqdm import tqdm

human_files_short = human_files[:100]
dog_files_short = dog_files[:100]

#-#-# Do NOT modify the code above this line. #-#-#

## TODO: Test the performance of the face_detector algorithm 
## on the images in human_files_short and dog_files_short.

humans = 0
dogs = 0

for h in human_files_short:
    if face_detector(h):
        humans += 1

print('Performance of human faces:',humans/len(human_files_short),'%')

for g in dog_files_short:
    if face_detector(g):
        dogs+=1
print('Performance of dog faces:',dogs/len(dog_files_short),'%')

        
Performance of human faces: 0.98 %
Performance of dog faces: 0.17 %

We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on human_files_short and dog_files_short.

In [5]:
### (Optional) 
### TODO: Test performance of anotherface detection algorithm.
### Feel free to use as many code cells as needed.

Step 2: Detect Dogs

In this section, we use a pre-trained model to detect dogs in images.

Obtain Pre-trained VGG-16 Model

The code cell below downloads the VGG-16 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories.

In [6]:
import torch
import torchvision.models as models

# define VGG16 model
VGG16 = models.vgg16(pretrained=True)

# check if CUDA is available
use_cuda = torch.cuda.is_available()

# move model to GPU if CUDA is available
if use_cuda:
    VGG16 = VGG16.cuda()
Downloading: "https://download.pytorch.org/models/vgg16-397923af.pth" to /root/.torch/models/vgg16-397923af.pth
100%|██████████| 553433881/553433881 [00:30<00:00, 17985128.89it/s]

Given an image, this pre-trained VGG-16 model returns a prediction (derived from the 1000 possible categories in ImageNet) for the object that is contained in the image.

(IMPLEMENTATION) Making Predictions with a Pre-trained Model

In the next code cell, you will write a function that accepts a path to an image (such as 'dogImages/train/001.Affenpinscher/Affenpinscher_00001.jpg') as input and returns the index corresponding to the ImageNet class that is predicted by the pre-trained VGG-16 model. The output should always be an integer between 0 and 999, inclusive.

Before writing the function, make sure that you take the time to learn how to appropriately pre-process tensors for pre-trained models in the PyTorch documentation.

In [7]:
from PIL import Image
import torchvision.transforms as transforms

def VGG16_predict(img_path):
    '''
    Use pre-trained VGG-16 model to obtain index corresponding to 
    predicted ImageNet class for image at specified path
    
    Args:
        img_path: path to an image
        
    Returns:
        Index corresponding to VGG-16 model's prediction
    '''
    
    ## TODO: Complete the function.
    ## Load and pre-process an image from the given img_path
    ## Return the *index* of the predicted class for that image
    
    img = Image.open(img_path)
    
    trans = transforms.Compose([transforms.Resize(256),
                                    transforms.CenterCrop(size=224),
                                    transforms.ToTensor(),
                                    transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                         std=[0.229, 0.224, 0.225])]) # creating the normalize transformation
    
    img_trans = trans(img) #apply the transform
    
    
    batch_t = img_trans.unsqueeze(0)
    batch_t = batch_t.to('cuda')
    
    
    VGG16.eval() #eval mode
    
    ## model inference
    out = VGG16(batch_t)
    out = out.to('cpu')
    
    out = out.data.numpy().argmax()  # prediction
    
    return out # predicted class index
In [8]:
VGG16_predict(dog_files_short[0])
Out[8]:
243

(IMPLEMENTATION) Write a Dog Detector

While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained VGG-16 model, we need only check if the pre-trained model predicts an index between 151 and 268 (inclusive).

Use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).

In [9]:
### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
    ## TODO: Complete the function.
    if VGG16_predict(img_path) in range(151, 269):
        return True
    else:
        return False

(IMPLEMENTATION) Assess the Dog Detector

Question 2: Use the code cell below to test the performance of your dog_detector function.

  • What percentage of the images in human_files_short have a detected dog?
  • What percentage of the images in dog_files_short have a detected dog?

Answer:

What percentage of the images in human_files_short have a detected dog?
In [10]:
### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.
human = 0

for h in human_files_short:
    if dog_detector(h):
        human += 1
        
print('Percentage of human in dog detector is:',human/len(human_files_short))
Percentage of human in dog detector is: 0.0
What percentage of the images in dog_files_short have a detected dog?
In [11]:
dog = 0

for g in dog_files_short:
    if dog_detector(g):
        dog += 1
        
print('Percentage of dog in dog detector is:',dog/len(dog_files_short))
Percentage of dog in dog detector is: 1.0

100% is very great

We suggest VGG-16 as a potential network to detect dog images in your algorithm, but you are free to explore other pre-trained networks (such as Inception-v3, ResNet-50, etc). Please use the code cell below to test other pre-trained PyTorch models. If you decide to pursue this optional task, report performance on human_files_short and dog_files_short.

In [12]:
### (Optional) 
### TODO: Report the performance of another pre-trained network.
### Feel free to use as many code cells as needed.

Step 3: Create a CNN to Classify Dog Breeds (from Scratch)

Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 10%. In Step 4 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.

We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have trouble distinguishing between a Brittany and a Welsh Springer Spaniel.

Brittany Welsh Springer Spaniel

It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).

Curly-Coated Retriever American Water Spaniel

Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.

Yellow Labrador Chocolate Labrador Black Labrador

We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.

Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!

(IMPLEMENTATION) Specify Data Loaders for the Dog Dataset

Use the code cell below to write three separate data loaders for the training, validation, and test datasets of dog images (located at dog_images/train, dog_images/valid, and dog_images/test, respectively). You may find this documentation on custom datasets to be a useful resource. If you are interested in augmenting your training and/or validation data, check out the wide variety of transforms!

In [13]:
import os
import torchvision.transforms as transforms
from torchvision import datasets
from PIL import ImageFile

ImageFile.LOAD_TRUNCATED_IMAGES = True

### TODO: Write data loaders for training, validation, and test sets
## Specify appropriate transforms, and batch_sizes

transforms = {
    
    # RandomHorizontalFlip() & RandomRotation() to augement data in train transformation
    'train' : transforms.Compose([transforms.Resize(256),
                                  transforms.RandomResizedCrop(224),
                                  transforms.RandomHorizontalFlip(),
                                  transforms.RandomRotation(10),
                                  transforms.ToTensor(),
                                  transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                       std=[0.229, 0.224, 0.225])]),
    
    'valid' : transforms.Compose([transforms.Resize(256),
                                  transforms.CenterCrop(224),
                                  transforms.ToTensor(),
                                  transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                       std=[0.229, 0.224, 0.225])]),
    
    'test' : transforms.Compose([transforms.Resize(256),
                                 transforms.CenterCrop(224),
                                 transforms.ToTensor(),
                                 transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                      std=[0.229, 0.224, 0.225])])
}

num_workers = 0
batch_size = 20

image_datasets = {x: datasets.ImageFolder(os.path.join('/data/dog_images', x), transforms[x])
                 for x in ['train', 'valid', 'test']}

data_loaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=batch_size,
                                              shuffle=True, num_workers=num_workers)
               for x in ['train', 'valid', 'test']}
In [14]:
from torchvision import utils

def visualize_sample_images(inp):
    inp = inp.numpy().transpose((1, 2, 0))
    inp = inp * np.array((0.229, 0.224, 0.225)) + np.array((0.485, 0.456, 0.406))
    inp = np.clip(inp, 0, 1)
    
    fig = plt.figure(figsize=(60, 25))
    plt.axis('off')
    plt.imshow(inp)
    plt.pause(0.001)
    
# Get a batch of training data.    
inputs, classes = next(iter(data_loaders['train']))

# Convert the batch to a grid.
grid = utils.make_grid(inputs, nrow=5)

# Display!
visualize_sample_images(grid)

Question 3: Describe your chosen procedure for preprocessing the data.

  • How does your code resize the images (by cropping, stretching, etc)? What size did you pick for the input tensor, and why?
  • Did you decide to augment the dataset? If so, how (through translations, flips, rotations, etc)? If not, why not?

Answer:

  • Changing the size to 256, then cropping to 224x224 pixels, since reducing the size of the images could greatly increase the execution time.

  • Yes, arguing the data set adds more value to training as it trains with flips and rotations.

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. Use the template in the code cell below.

In [15]:
import torch.nn as nn
import torch.nn.functional as F

# define the CNN architecture
class Net(nn.Module):
    ### TODO: choose an architecture, and complete the class
    def __init__(self):
        super(Net, self).__init__()
        ## Define layers of a CNN
        self.conv1 = nn.Conv2d(3, 32, 3, stride=2, padding=1)      
        self.conv2 = nn.Conv2d(32, 64, 3, stride=2, padding=1)
        self.conv3 = nn.Conv2d(64, 128, 3, padding=1)
        self.pool = nn.MaxPool2d(2, 2)
        self.fc1 = nn.Linear(7*7*128, 512)
        self.fc2 = nn.Linear(512, len(image_datasets['train'].classes))
        self.dropout = nn.Dropout(0.25)
    
    def forward(self, x):
        ## Define forward behavior
        x = F.relu(self.conv1(x)) # shape 112
        x = self.pool(x) # shape 56
        x = F.relu(self.conv2(x)) # shape 28
        x = self.pool(x) # size 14
        x = F.relu(self.conv3(x)) # shape 14
        x = self.pool(x) # shape 7
        
        x = x.view(x.size(0), -1)
        
        x = self.dropout(x)
        x = F.relu(self.fc1(x))
        
        x = self.dropout(x)
        x = self.fc2(x)
        return x

#-#-# You so NOT have to modify the code below this line. #-#-#

# instantiate the CNN
model_scratch = Net()

# move tensors to GPU if CUDA is available
if use_cuda:
    model_scratch.cuda()

Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step.

Answer:

First input layer has shape (224, 224, 3) while the last layer have 133 output classes.

(input): (shape: 224, 224, 3) (conv1): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1)) with ReLU activation (shape: 112, 112, 3) (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False) (shape: 56, 56, 3) (conv2): Conv2d(32, 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1)) with ReLU activation (shape: 28, 28, 3) (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False) (shape: 14, 14, 3) (conv3): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) with ReLU activation (shape: 14, 14, 3) (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False) (shape: 7, 7, 3) (dropout): Dropout(p=0.25) (fc1): Linear(in_features=6272, out_features=512, bias=True) (dropout): Dropout(p=0.25) (fc2): Linear(in_features=512, out_features=133, bias=True)

Convolution layers are used to extract features from an input image. The more convolutional layers there are, the more complex patterns in color and shape a model can detect. But to keep things simple only 3 convolution layers are used in the model. Maxpooling is added to downsample by a factor of 2 after each convolution layers. 25% dropout added before each fully-connected layer to prevent overfitting. Last fully-connected layer will produce the final output_size (133 dimension) which predicts the different classes of breeds.

(IMPLEMENTATION) Specify Loss Function and Optimizer

Use the next code cell to specify a loss function and optimizer. Save the chosen loss function as criterion_scratch, and the optimizer as optimizer_scratch below.

In [16]:
import torch.optim as optim

### TODO: select loss function
criterion_scratch = nn.CrossEntropyLoss()

### TODO: select optimizer
optimizer_scratch = optim.SGD(model_scratch.parameters(), lr=0.01)

(IMPLEMENTATION) Train and Validate the Model

Train and validate your model in the code cell below. Save the final model parameters at filepath 'model_scratch.pt'.

In [17]:
def train(n_epochs, loaders, model, optimizer, criterion, use_cuda, save_path):
    """returns trained model"""
    # initialize tracker for minimum validation loss
    valid_loss_min = np.Inf 
    
    for epoch in range(1, n_epochs+1):
        # initialize variables to monitor training and validation loss
        train_loss = 0.0
        valid_loss = 0.0
        
        ###################
        # train the model #
        ###################
        model.train()
        for batch_idx, (data, target) in enumerate(loaders['train']):
            # move to GPU
            if use_cuda:
                data, target = data.cuda(), target.cuda()
            ## find the loss and update the model parameters accordingly
            ## record the average training loss, using something like
            ## train_loss = train_loss + ((1 / (batch_idx + 1)) * (loss.data - train_loss))
            optimizer.zero_grad()
            # forward pass
            output = model(data)
            # calculate batch loss
            loss = criterion(output, target)
            # backward pass
            loss.backward()
            # parameter update
            optimizer.step()
            # update training loss
            train_loss += loss.item() * data.size(0)
            
        ######################    
        # validate the model #
        ######################
        model.eval()
        for batch_idx, (data, target) in enumerate(loaders['valid']):
            # move to GPU
            if use_cuda:
                data, target = data.cuda(), target.cuda()
            ## update the average validation loss
            # forward pass
            output = model(data)
            # batch loss
            loss = criterion(output, target)
            # update validation loss
            valid_loss += loss.item() * data.size(0)
            
        # calculate average losses
        train_loss = train_loss/len(loaders['train'].dataset)
        valid_loss = valid_loss/len(loaders['valid'].dataset)

            
        # print training/validation statistics 
        print('Epoch: {} \tTraining Loss: {:.6f} \tValidation Loss: {:.6f}'.format(
            epoch, 
            train_loss,
            valid_loss
            ))
        
        ## TODO: save the model if validation loss has decreased
        if valid_loss <= valid_loss_min:
            print('Validation loss decreased ({:.6f} --> {:.6f}).    Saving model...'.
                 format(valid_loss_min, valid_loss))
            torch.save(model.state_dict(), save_path)
            valid_loss_min = valid_loss
            
    # return trained model
    return model


# train the model
model_scratch = train(100, data_loaders, model_scratch, optimizer_scratch, 
                      criterion_scratch, use_cuda, 'model_scratch.pt')

# load the model that got the best validation accuracy
model_scratch.load_state_dict(torch.load('model_scratch.pt'))
Epoch: 1 	Training Loss: 4.881630 	Validation Loss: 4.865662
Validation loss decreased (inf --> 4.865662).    Saving model...
Epoch: 2 	Training Loss: 4.855406 	Validation Loss: 4.823151
Validation loss decreased (4.865662 --> 4.823151).    Saving model...
Epoch: 3 	Training Loss: 4.807092 	Validation Loss: 4.736662
Validation loss decreased (4.823151 --> 4.736662).    Saving model...
Epoch: 4 	Training Loss: 4.757834 	Validation Loss: 4.682348
Validation loss decreased (4.736662 --> 4.682348).    Saving model...
Epoch: 5 	Training Loss: 4.730545 	Validation Loss: 4.645934
Validation loss decreased (4.682348 --> 4.645934).    Saving model...
Epoch: 6 	Training Loss: 4.682821 	Validation Loss: 4.568208
Validation loss decreased (4.645934 --> 4.568208).    Saving model...
Epoch: 7 	Training Loss: 4.610359 	Validation Loss: 4.486095
Validation loss decreased (4.568208 --> 4.486095).    Saving model...
Epoch: 8 	Training Loss: 4.554565 	Validation Loss: 4.425702
Validation loss decreased (4.486095 --> 4.425702).    Saving model...
Epoch: 9 	Training Loss: 4.531876 	Validation Loss: 4.400605
Validation loss decreased (4.425702 --> 4.400605).    Saving model...
Epoch: 10 	Training Loss: 4.527200 	Validation Loss: 4.371331
Validation loss decreased (4.400605 --> 4.371331).    Saving model...
Epoch: 11 	Training Loss: 4.489069 	Validation Loss: 4.335948
Validation loss decreased (4.371331 --> 4.335948).    Saving model...
Epoch: 12 	Training Loss: 4.471941 	Validation Loss: 4.314975
Validation loss decreased (4.335948 --> 4.314975).    Saving model...
Epoch: 13 	Training Loss: 4.451627 	Validation Loss: 4.287775
Validation loss decreased (4.314975 --> 4.287775).    Saving model...
Epoch: 14 	Training Loss: 4.420789 	Validation Loss: 4.264119
Validation loss decreased (4.287775 --> 4.264119).    Saving model...
Epoch: 15 	Training Loss: 4.378053 	Validation Loss: 4.201215
Validation loss decreased (4.264119 --> 4.201215).    Saving model...
Epoch: 16 	Training Loss: 4.359790 	Validation Loss: 4.275794
Epoch: 17 	Training Loss: 4.344786 	Validation Loss: 4.158182
Validation loss decreased (4.201215 --> 4.158182).    Saving model...
Epoch: 18 	Training Loss: 4.324204 	Validation Loss: 4.138976
Validation loss decreased (4.158182 --> 4.138976).    Saving model...
Epoch: 19 	Training Loss: 4.306277 	Validation Loss: 4.146388
Epoch: 20 	Training Loss: 4.283808 	Validation Loss: 4.100685
Validation loss decreased (4.138976 --> 4.100685).    Saving model...
Epoch: 21 	Training Loss: 4.260341 	Validation Loss: 4.089696
Validation loss decreased (4.100685 --> 4.089696).    Saving model...
Epoch: 22 	Training Loss: 4.231360 	Validation Loss: 4.066221
Validation loss decreased (4.089696 --> 4.066221).    Saving model...
Epoch: 23 	Training Loss: 4.209670 	Validation Loss: 4.184201
Epoch: 24 	Training Loss: 4.223814 	Validation Loss: 4.089601
Epoch: 25 	Training Loss: 4.189141 	Validation Loss: 4.004647
Validation loss decreased (4.066221 --> 4.004647).    Saving model...
Epoch: 26 	Training Loss: 4.149329 	Validation Loss: 4.068530
Epoch: 27 	Training Loss: 4.129512 	Validation Loss: 3.953040
Validation loss decreased (4.004647 --> 3.953040).    Saving model...
Epoch: 28 	Training Loss: 4.118536 	Validation Loss: 3.944937
Validation loss decreased (3.953040 --> 3.944937).    Saving model...
Epoch: 29 	Training Loss: 4.090003 	Validation Loss: 3.935095
Validation loss decreased (3.944937 --> 3.935095).    Saving model...
Epoch: 30 	Training Loss: 4.059526 	Validation Loss: 3.957703
Epoch: 31 	Training Loss: 4.038920 	Validation Loss: 3.885381
Validation loss decreased (3.935095 --> 3.885381).    Saving model...
Epoch: 32 	Training Loss: 4.029705 	Validation Loss: 3.892090
Epoch: 33 	Training Loss: 4.004112 	Validation Loss: 3.848600
Validation loss decreased (3.885381 --> 3.848600).    Saving model...
Epoch: 34 	Training Loss: 3.961889 	Validation Loss: 3.846673
Validation loss decreased (3.848600 --> 3.846673).    Saving model...
Epoch: 35 	Training Loss: 3.954432 	Validation Loss: 3.860166
Epoch: 36 	Training Loss: 3.928279 	Validation Loss: 3.832251
Validation loss decreased (3.846673 --> 3.832251).    Saving model...
Epoch: 37 	Training Loss: 3.916264 	Validation Loss: 3.806691
Validation loss decreased (3.832251 --> 3.806691).    Saving model...
Epoch: 38 	Training Loss: 3.884460 	Validation Loss: 3.764061
Validation loss decreased (3.806691 --> 3.764061).    Saving model...
Epoch: 39 	Training Loss: 3.868308 	Validation Loss: 3.809322
Epoch: 40 	Training Loss: 3.842515 	Validation Loss: 3.794831
Epoch: 41 	Training Loss: 3.798259 	Validation Loss: 3.827599
Epoch: 42 	Training Loss: 3.796024 	Validation Loss: 3.804844
Epoch: 43 	Training Loss: 3.790025 	Validation Loss: 3.767132
Epoch: 44 	Training Loss: 3.756572 	Validation Loss: 3.739425
Validation loss decreased (3.764061 --> 3.739425).    Saving model...
Epoch: 45 	Training Loss: 3.729859 	Validation Loss: 3.705995
Validation loss decreased (3.739425 --> 3.705995).    Saving model...
Epoch: 46 	Training Loss: 3.690893 	Validation Loss: 3.736487
Epoch: 47 	Training Loss: 3.694108 	Validation Loss: 3.701707
Validation loss decreased (3.705995 --> 3.701707).    Saving model...
Epoch: 48 	Training Loss: 3.663949 	Validation Loss: 3.679517
Validation loss decreased (3.701707 --> 3.679517).    Saving model...
Epoch: 49 	Training Loss: 3.661618 	Validation Loss: 3.693940
Epoch: 50 	Training Loss: 3.638475 	Validation Loss: 3.654665
Validation loss decreased (3.679517 --> 3.654665).    Saving model...
Epoch: 51 	Training Loss: 3.611179 	Validation Loss: 3.664147
Epoch: 52 	Training Loss: 3.598067 	Validation Loss: 3.734970
Epoch: 53 	Training Loss: 3.570981 	Validation Loss: 3.640314
Validation loss decreased (3.654665 --> 3.640314).    Saving model...
Epoch: 54 	Training Loss: 3.547240 	Validation Loss: 3.674075
Epoch: 55 	Training Loss: 3.520964 	Validation Loss: 3.575421
Validation loss decreased (3.640314 --> 3.575421).    Saving model...
Epoch: 56 	Training Loss: 3.507171 	Validation Loss: 3.583061
Epoch: 57 	Training Loss: 3.479482 	Validation Loss: 3.563229
Validation loss decreased (3.575421 --> 3.563229).    Saving model...
Epoch: 58 	Training Loss: 3.481219 	Validation Loss: 3.611533
Epoch: 59 	Training Loss: 3.439140 	Validation Loss: 3.553581
Validation loss decreased (3.563229 --> 3.553581).    Saving model...
Epoch: 60 	Training Loss: 3.426809 	Validation Loss: 3.568734
Epoch: 61 	Training Loss: 3.399536 	Validation Loss: 3.596151
Epoch: 62 	Training Loss: 3.380545 	Validation Loss: 3.590739
Epoch: 63 	Training Loss: 3.378784 	Validation Loss: 3.626261
Epoch: 64 	Training Loss: 3.376139 	Validation Loss: 3.571355
Epoch: 65 	Training Loss: 3.340934 	Validation Loss: 3.608213
Epoch: 66 	Training Loss: 3.314913 	Validation Loss: 3.555949
Epoch: 67 	Training Loss: 3.303941 	Validation Loss: 3.521583
Validation loss decreased (3.553581 --> 3.521583).    Saving model...
Epoch: 68 	Training Loss: 3.309206 	Validation Loss: 3.516459
Validation loss decreased (3.521583 --> 3.516459).    Saving model...
Epoch: 69 	Training Loss: 3.264474 	Validation Loss: 3.447297
Validation loss decreased (3.516459 --> 3.447297).    Saving model...
Epoch: 70 	Training Loss: 3.253451 	Validation Loss: 3.509477
Epoch: 71 	Training Loss: 3.214004 	Validation Loss: 3.575995
Epoch: 72 	Training Loss: 3.204194 	Validation Loss: 3.556703
Epoch: 73 	Training Loss: 3.184636 	Validation Loss: 3.549562
Epoch: 74 	Training Loss: 3.174317 	Validation Loss: 3.513124
Epoch: 75 	Training Loss: 3.170278 	Validation Loss: 3.470485
Epoch: 76 	Training Loss: 3.152153 	Validation Loss: 3.508343
Epoch: 77 	Training Loss: 3.119316 	Validation Loss: 3.477614
Epoch: 78 	Training Loss: 3.101027 	Validation Loss: 3.527302
Epoch: 79 	Training Loss: 3.079822 	Validation Loss: 3.472771
Epoch: 80 	Training Loss: 3.082753 	Validation Loss: 3.483573
Epoch: 81 	Training Loss: 3.063476 	Validation Loss: 3.494451
Epoch: 82 	Training Loss: 3.025187 	Validation Loss: 3.568223
Epoch: 83 	Training Loss: 3.038407 	Validation Loss: 3.497779
Epoch: 84 	Training Loss: 2.995128 	Validation Loss: 3.532985
Epoch: 85 	Training Loss: 2.992526 	Validation Loss: 3.541097
Epoch: 86 	Training Loss: 2.969015 	Validation Loss: 3.595415
Epoch: 87 	Training Loss: 2.966222 	Validation Loss: 3.553086
Epoch: 88 	Training Loss: 2.955441 	Validation Loss: 3.620909
Epoch: 89 	Training Loss: 2.950137 	Validation Loss: 3.518595
Epoch: 90 	Training Loss: 2.913636 	Validation Loss: 3.484224
Epoch: 91 	Training Loss: 2.895953 	Validation Loss: 3.522604
Epoch: 92 	Training Loss: 2.907341 	Validation Loss: 3.614597
Epoch: 93 	Training Loss: 2.863694 	Validation Loss: 3.520199
Epoch: 94 	Training Loss: 2.867273 	Validation Loss: 3.533733
Epoch: 95 	Training Loss: 2.817854 	Validation Loss: 3.618249
Epoch: 96 	Training Loss: 2.798091 	Validation Loss: 3.591803
Epoch: 97 	Training Loss: 2.811839 	Validation Loss: 3.583822
Epoch: 98 	Training Loss: 2.762218 	Validation Loss: 3.592567
Epoch: 99 	Training Loss: 2.760919 	Validation Loss: 3.636280
Epoch: 100 	Training Loss: 2.780101 	Validation Loss: 3.511796

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Use the code cell below to calculate and print the test loss and accuracy. Ensure that your test accuracy is greater than 10%.

In [18]:
def test(loaders, model, criterion, use_cuda):

    # monitor test loss and accuracy
    test_loss = 0.
    correct = 0.
    total = 0.

    model.eval()
    for batch_idx, (data, target) in enumerate(loaders['test']):
        # move to GPU
        if use_cuda:
            data, target = data.cuda(), target.cuda()
        # forward pass: compute predicted outputs by passing inputs to the model
        output = model(data)
        # calculate the loss
        loss = criterion(output, target)
        # update average test loss 
        test_loss = test_loss + ((1 / (batch_idx + 1)) * (loss.data - test_loss))
        # convert output probabilities to predicted class
        pred = output.data.max(1, keepdim=True)[1]
        # compare predictions to true label
        correct += np.sum(np.squeeze(pred.eq(target.data.view_as(pred))).cpu().numpy())
        total += data.size(0)
            
    print('Test Loss: {:.6f}\n'.format(test_loss))

    print('\nTest Accuracy: %2d%% (%2d/%2d)' % (
        100. * correct / total, correct, total))

# call test function    
test(data_loaders, model_scratch, criterion_scratch, use_cuda)
Test Loss: 3.425855


Test Accuracy: 19% (167/836)

Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)

You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.

(IMPLEMENTATION) Specify Data Loaders for the Dog Dataset

Use the code cell below to write three separate data loaders for the training, validation, and test datasets of dog images (located at dogImages/train, dogImages/valid, and dogImages/test, respectively).

If you like, you are welcome to use the same data loaders from the previous step, when you created a CNN from scratch.

In [19]:
## TODO: Specify data loaders
transfer_loaders = data_loaders.copy()

(IMPLEMENTATION) Model Architecture

Use transfer learning to create a CNN to classify dog breed. Use the code cell below, and save your initialized model as the variable model_transfer.

In [20]:
import torchvision.models as models
import torch.nn as nn

## TODO: Specify model architecture 
model_transfer = models.resnet50(pretrained=True)

for param in model_transfer.parameters():
    param.requires_grad = False
    
model_transfer.fc = nn.Linear(2048, 133)



if use_cuda:
    model_transfer = model_transfer.cuda()
Downloading: "https://download.pytorch.org/models/resnet50-19c8e357.pth" to /root/.torch/models/resnet50-19c8e357.pth
100%|██████████| 102502400/102502400 [00:00<00:00, 104401528.78it/s]

Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.

Answer:

Using pre-trained models such as ResNet50 is a quick and efficient way to handle computer vision problems. As ResNet was trained on ImageNet for classification on a wide range of objects, a transfer learning approach is required to work in our problem set. We could make use of the pretrained features by freezing the layers and changing the last fully connected layer to predict 133 classes (different dog breeds in this problem set)

Choice of criterion: nn.CrossEntropyLoss() This criterion combines :func:nn.LogSoftmax and :func:nn.NLLLoss in one single class. It is useful when training a classification problem with C classes.

(IMPLEMENTATION) Specify Loss Function and Optimizer

Use the next code cell to specify a loss function and optimizer. Save the chosen loss function as criterion_transfer, and the optimizer as optimizer_transfer below.

In [21]:
criterion_transfer = nn.CrossEntropyLoss()
optimizer_transfer = optim.Adam(model_transfer.fc.parameters(), lr=0.001)

(IMPLEMENTATION) Train and Validate the Model

Train and validate your model in the code cell below. Save the final model parameters at filepath 'model_transfer.pt'.

In [22]:
# train the model
model_transfer =  train(10, transfer_loaders, model_transfer, optimizer_transfer, criterion_transfer, use_cuda, 'model_transfer.pt')
# load the model that got the best validation accuracy (uncomment the line below)
model_transfer.load_state_dict(torch.load('model_transfer.pt'))
Epoch: 1 	Training Loss: 2.839860 	Validation Loss: 0.991250
Validation loss decreased (inf --> 0.991250).    Saving model...
Epoch: 2 	Training Loss: 1.518533 	Validation Loss: 0.711634
Validation loss decreased (0.991250 --> 0.711634).    Saving model...
Epoch: 3 	Training Loss: 1.384541 	Validation Loss: 0.652656
Validation loss decreased (0.711634 --> 0.652656).    Saving model...
Epoch: 4 	Training Loss: 1.311629 	Validation Loss: 0.634810
Validation loss decreased (0.652656 --> 0.634810).    Saving model...
Epoch: 5 	Training Loss: 1.228826 	Validation Loss: 0.610240
Validation loss decreased (0.634810 --> 0.610240).    Saving model...
Epoch: 6 	Training Loss: 1.172907 	Validation Loss: 0.624041
Epoch: 7 	Training Loss: 1.167559 	Validation Loss: 0.579296
Validation loss decreased (0.610240 --> 0.579296).    Saving model...
Epoch: 8 	Training Loss: 1.171670 	Validation Loss: 0.611298
Epoch: 9 	Training Loss: 1.110848 	Validation Loss: 0.529020
Validation loss decreased (0.579296 --> 0.529020).    Saving model...
Epoch: 10 	Training Loss: 1.097893 	Validation Loss: 0.484041
Validation loss decreased (0.529020 --> 0.484041).    Saving model...

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Use the code cell below to calculate and print the test loss and accuracy. Ensure that your test accuracy is greater than 60%.

In [24]:
test(transfer_loaders, model_transfer, criterion_transfer, use_cuda)
Test Loss: 0.504100


Test Accuracy: 85% (714/836)

(IMPLEMENTATION) Predict Dog Breed with the Model

Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan hound, etc) that is predicted by your model.

In [28]:
### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.

from PIL import Image
import torchvision.transforms as transforms

class_names = [item[4:].replace("_", " ") for item in transfer_loaders['train'].dataset.classes]

def load_input_image(img_path):    
    image = Image.open(img_path).convert('RGB')
    prediction_transform = transforms.Compose([transforms.Resize(size=(224, 224)),
                                               transforms.ToTensor(), 
                                               transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                                    std=[0.229, 0.224, 0.225])])

    # discard the transparent, alpha channel (that's the :3) and add the batch dimension
    image = prediction_transform(image)[:3,:,:].unsqueeze(0)
    return image

def predict_breed_transfer(model, class_names, img_path):
    # load the image and return the predicted breed
    img = load_input_image(img_path)
    model = model.cpu()
    model.eval()
    idx = torch.argmax(model(img))
    return class_names[idx]

for img_file in os.listdir('./images'):
    img_path = os.path.join('./images', img_file)
    predition = predict_breed_transfer(model_transfer, class_names, img_path)
    print("image_file_name: {0}, \t predition breed: {1}".format(img_path, predition))
image_file_name: ./images/Labrador_retriever_06449.jpg, 	 predition breed: Labrador retriever
image_file_name: ./images/sample_dog_output.png, 	 predition breed: Entlebucher mountain dog
image_file_name: ./images/Brittany_02625.jpg, 	 predition breed: Brittany
image_file_name: ./images/sample_human_output.png, 	 predition breed: Bulldog
image_file_name: ./images/American_water_spaniel_00648.jpg, 	 predition breed: Curly-coated retriever
image_file_name: ./images/Curly-coated_retriever_03896.jpg, 	 predition breed: Curly-coated retriever
image_file_name: ./images/Labrador_retriever_06457.jpg, 	 predition breed: Labrador retriever
image_file_name: ./images/Labrador_retriever_06455.jpg, 	 predition breed: Labrador retriever
image_file_name: ./images/Welsh_springer_spaniel_08203.jpg, 	 predition breed: Irish red and white setter
image_file_name: ./images/sample_cnn.png, 	 predition breed: American eskimo dog

Step 5: Write your Algorithm

Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,

  • if a dog is detected in the image, return the predicted breed.
  • if a human is detected in the image, return the resembling dog breed.
  • if neither is detected in the image, provide output that indicates an error.

You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and human_detector functions developed above. You are required to use your CNN from Step 4 to predict dog breed.

Some sample output for our algorithm is provided below, but feel free to design your own user experience!

Sample Human Output

(IMPLEMENTATION) Write your Algorithm

In [31]:
### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.

def run_app(img_path):
    ## handle cases for a human face, dog, and neither
    img = Image.open(img_path)
    plt.imshow(img)
    plt.show()
    
    if dog_detector(img_path) is True:
        prediction = predict_breed_transfer(model_transfer, class_names, img_path)
        print("Dog Detected!\nIt looks like a {0}".format(prediction))  
    elif face_detector(img_path) > 0:
        prediction = predict_breed_transfer(model_transfer, class_names, img_path)
        print("You look like a ... \n {0}".format(prediction))
    else:
        print("sorry i dont know ?¿")

Step 6: Test Your Algorithm

In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?

(IMPLEMENTATION) Test Your Algorithm on Sample Images!

Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.

Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.

Answer: (Three possible points for improvement)

  • Using transfer learning of a model that is similar to the one being studied can increase performance effectiveness

  • Variations in the images are likely to increase the effectiveness of the algorithm.

  • Knowing the classes that will be classified increases the effectiveness in predicting classes

In [38]:
import random
In [44]:
l = [random.randint(0,100) for i in range(15)]
In [46]:
## TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.

## suggested code, below
for file in np.hstack((human_files[l], dog_files[l])):
    run_app(file)
You look like a ... 
 Bearded collie
You look like a ... 
 Bullmastiff
You look like a ... 
 Entlebucher mountain dog
You look like a ... 
 Norfolk terrier
You look like a ... 
 English springer spaniel
You look like a ... 
 Norfolk terrier
You look like a ... 
 Newfoundland
You look like a ... 
 Glen of imaal terrier
You look like a ... 
 Chihuahua
You look like a ... 
 Dachshund
You look like a ... 
 Poodle
You look like a ... 
 Poodle
You look like a ... 
 Glen of imaal terrier
You look like a ... 
 Xoloitzcuintli
You look like a ... 
 Chinese shar-pei
Dog Detected!
It looks like a Mastiff
Dog Detected!
It looks like a Doberman pinscher
Dog Detected!
It looks like a Mastiff
Dog Detected!
It looks like a Great dane
Dog Detected!
It looks like a Mastiff
Dog Detected!
It looks like a Mastiff
Dog Detected!
It looks like a Doberman pinscher
Dog Detected!
It looks like a Mastiff
Dog Detected!
It looks like a Bullmastiff
Dog Detected!
It looks like a Doberman pinscher
Dog Detected!
It looks like a German pinscher
Dog Detected!
It looks like a Mastiff
Dog Detected!
It looks like a Mastiff
Dog Detected!
It looks like a German pinscher
Dog Detected!
It looks like a Doberman pinscher
In [ ]: